The effects of water table draw-down on the hydrology of a patterned fen peatland near Quebec City, Quebec, Canada. Peter Nicholas Whittington

Transcription

1 The effects of water table draw-down on the hydrology of a patterned fen peatland near Quebec City, Quebec, Canada by Peter Nicholas Whittington A thesis presented to the University of Waterloo in fulfilment of the thesis requirement for the degree of Master of Environmental Studies in Geography Waterloo, Ontario, Canada, 2005 Peter Nicholas Whittington 2005

2 AUTHOR S DECLARATION FOR ELECTRONIC SUBMISSION OF A THESIS I hereby declare that I am the sole author of this thesis. This is a true copy of the thesis, including any required final revisions, as accepted by my examiners. I understand that my thesis may be made electronically available to the public. ii

3 Abstract Hydrological response to climate change may alter the biogeochemical role that peatlands play in the global climate system, so an understanding of the nature and magnitude of this response is important. Simple hydrological models have predicted the effects of climate change on the hydrology of these systems, and estimated a ~20 cm water table draw down. This draw down amount was modeled to estimate the changing role that wetlands may play in global biogeochemical cycling, but failed to account for modifications of the peatland structure, which has profound implications for the hydrology of these systems. Volume change in compressible soils occurs as the result of different processes, mainly compression and oxidation. Compression occurs instantaneously as a change in water pressure (e.g., from water table draw down) occurs and the peat matrix is unable to withstand the increased pressures and subsides. Oxidation is the long term chemical breakdown of the peat under aerobic conditions. Consequently, in 2002 the water table in a fen peatland near Quebec City was lowered by ~20 cm (Experimental site), and the hydrological response was measured compared to a Control (no manipulation) and Drained site (previously drained c. 1994). As a result of the draw-down, the surface in the Experimental pool decreased 5, 15 and 20 cm in the ridge, lawn and mat topographic locations, respectively resulting in an increased bulk density of ~60% in the Experimental lawn. Hydraulic conductivity (K) generally decreased with depth and from Control (25 to 125 cm) 10-1 to 10-5 cm s -1 to Experimental (25 to 125 cm) 10-2 to 10-7 cm s -1 and to Drained (25 to 75 cm) 10-2 to 10-6 cm s -1. In similar iii

4 topographic locations (ridge, lawn, mat), K trended Control>Experimental>Drained, usually by an order of magnitude. Water table fluctuations in the Drained site were, on average, twice that of the Control site, whereas water table fluctuations within sites trended ridge>lawn>mat. The water table in the Control lawn was able to remain at a stable depth relative to the surface (~ -1 cm) because the lawn peat floats with changes in water table position. However, because of the denser, degraded peat, the Drained lawn peat was more rigid, forcing the water to fluctuate relative to the surface, further enhancing peat decay and densification. While climatic change will not occur instantaneously the limitations of the experiment required an abrupt change in water table position (drainage). However, regardless of how volume change occurs in the peat (compression or oxidation) the direction of change to the hydraulic properties is the same (increased bulk density, decreased hydraulic conductivity) which affects the hydrology of these systems (increase water table fluctuations and decreases surface movement). Thus, valuable information can be obtained regarding the changing role of peatlands in global biogeochemical cycling processes. iv

5 Acknowledgements I would like to thank my parents: my mom who got me interested in geography at a very early age; and my dad for giving me something to live up to. I thank my roommates, Adam and the other Peter, for providing me numerous opportunities for thesis procrastination. I especially thank Gareth Ward for help in the field. I hope your neck gets better I thank the McMaster crew, mainly Maria Strack, and Drs Mike Waddington, and Erik Kellner for help with vegetation identification, biogeochemical questions, and data analysis/collection, respectively. I also thank my girlfriend Rox for sticking with me, and for making me feel guilty for procrastinating writing my thesis; yet providing me with ample activities to the contrary. Lastly, I thank my advisor, Dr. Jonathan Price, without whom I would be standing on a frozen beach somewhere, rather than bogged down in a mosquito infested peatland. Among many things, you taught me that words should be weighed, not counted, and thus a simple thank you may do. v

6 Table of Contents 1. Introduction Context Patterned Peatlands: Formation and Function Formation Hydrology Peat volume change Effect of peat volume change on hydrological parameters Study Site Overview Nomenclature Methods Micro meteorological instrumentation Lysimeters Drainage Cores Soil Moisture Content Squishometers vi

7 3.7. Piezometer Installations and Locations Hydraulic Conductivity (K) Surface and Water Level Recorders Results Meteorological Hydrological Parameters Drainage Hydraulic Parameters Water Content Squishometers Hydraulic Conductivity Hydrology Water Table Position Surface Level and Water Table Fluctuations Mooratmung Discussion Meterological The Hydrology of Water Table Draw Down Long Term Water Table Change Hydrological Parameters vii

8 Seasonal Water Table and Surface Level Fluctuations The Positive Feedback Loop As a surrogate for climate change? Local Scale Hydrology of Patterned Peatlands Inter-pool Water Flow Intra-pool Water Flow Conclusion Appendix A Appendix B viii

9 List of Tables Table 3.1 Piezometer nest details and 2004 K test schedule, where 1, 2 and 3 refer to weeks 1, 2 and 3, respectively Table 3.2 Water and surface level recorder locations and years used. (sur = surface) Table 4.1 Average, minimum and maximum temperature values and precipitation totals per month for 2002 to 2004 seasons as well as the 30 year running average (*Environment Canada, 2005 at Quebec Lesage International Airport. Located 20 km west of the study site) Table 4.2 Actual (E a ) and Equilibrium (E q ) evaporation rates, and water deficit (P E) for 2002 to Table 4.3 Squishometer strain summary by layer. Values greater than 1 indicate swelling, and less than 1 compression Table 4.4 Surface elevation and water table elevation movements for 2003 and : Surface level recorders, 2: Water table recorders in topographic location, 3: Pool water table position (for Drained, the middle of the pool). Bold values represent the largest change within that category for a particular storm event. All values in cm except precipitation (mm) Table 5.1 Average discharge rates between pools based on average water table elevations and K values Table 5.2 Summary of specific discharge (q) for the average of K 50cm lawn and K 75cm lawn (Figure 4.8) and water table heights for wet and dry periods for 2002 to 2004 (based on Figure 4.13). dl is the linear distance between wells. Positive specific discharges indicate ridge to pool flow, whereas negative specific discharges represent pool to ridge flow. Areas were determined by multiplying the approximate perimeter distance by average peat depth. Perimeters were estimated to be 120, 40 and 80 m in C, E, and D, respectively, with peat depths of 1.4, 1.1, and.8 m, in C, E, and D, respectively.. 75 ix

10 List of Figures Figure 2.1 Ariel view of the Fen study site Figure 2.2 Site photos taken June 8, 2004 of the Control (top), Experimental (middle) and Drained (bottom) sites Figure 3.1 Site map of instrumentation Figure 3.2 Top: Shallower (5 cm) squishometer (upper) and deeper squishometer (lower) prior to installation; Bottom: Squishometers installed at the Experimental South site (note use of white measuring tape, opposed to yellow) Figure 3.3 Float-pulley system measuring the Control lawn water table Figure 4.1 Daily precipitation (positive black bars) and evaporation (negative black bars) for 2002 to 2004 with cumulative flux (P-E) (line). Evaporative data (meteorological station) was missing from JD in 2002 and thus the average daily loss was added to the cumulative line. Note different scale in negative direction for Figure 4.2 Frequency of 24 hour precipitation free periods for 2002 to 2004 field seasons. (In efforts to conserve the scale of < 24 hours (i.e., 1 on the y axis) is omitted because there are many (> 200) short time periods between rain events.) Figure 4.3 Surface subsidence along the Experimental transect. Yellow tape indicates the pre-drainage surface level with respect to the side of the piezometer Figure 4.4 Top: Surface subsidence measured manually relative to piezometer top for the Experimental pool, Bottom: Water table drawdown measurement manually relative to top of piezometer. Note: Drainage occurred on JD Figure 4.5 Bulk density (a) and specific Yield (b) in the Drained, Control, and Experimental lawns Figure 4.6 Soil Moisture Content (expressed as a proportion of total soil volume) for Depths of probes are indicated above/below the line. Water tables (W.T.) are indicated by C, E, or D W.T. For the first part of 2003, manual Experimental water tables are used (Man. E W.T.) Figure 4.7 Strain for squishometers Not all layers are shown as strain values are similar in magnitude. Note different vertical scales for all sites so that more detail could be shown Figure 4.8 Hydraulic Conductivity by topographic location within sites (a-c) and between sites (d-f). The dotted lines for Control in Figures a) and d) indicate that those K responses were too quick for manual measurement x

11 Figure 4.9 Hydraulic Conductivity through the CCZ and CEZ transects. The vertical line denotes the change between the CCZ and CEZ transects. The heavy dashed lines indicates that head recovery was too quick for manual measurement (50 cm) Figure 4.10 Hydraulic Conductivity through the EDZ transect Figure 4.11 Hydraulic Conductivity through the DDZ transect Figure 4.12 Weekly K values for the mat nests of the CEZ transect. (Julian day is the date of the Monday for the week the test was completed.) Figure 4.13 Hydraulic conductivity for the squishometer piezometers in 2002 and Note: vertical lines separate pre-drainage, 2002 and Figure 4.14 Water table location for wet and dry periods between pools for all years. Note: Dry and Wet are relative terms for each specific year. The four points along the X-axis represent, from left to right, ridge, lawn, mat, and pool locations. Note that the surface elevation changes decreases from Control to Drained. Surface is the peat surface in the ridge, lawn, mat and pool topographic location, at a dry and wet time, respectively. Water table is the water table in the ridge, lawn, mat and pool topographic location, at a dry and wet time, respectively Figure 4.15 Bi-hourly total precipitation and average water table elevation above a common datum, in pools. The large and rapid decline in the Experimental pool in 2002 (top) was the drainage of the Experimental pool Figure 4.16 Average ratio of change of hydrograph (mm) to precipitation (mm) input for 2003 (right) and 2004 (left). Bars are +/- standard deviation Figure 4.17 Lawn and Ridge (a and b) surface and water table elevations for the Control and Drained sites, Lawn and ridge (c and d) water table fluctuations, relative to the surface for the Drained and Control sites. (Note different vertical scales for c and d.) Figure 5.1 Conceptual diagram of water table draw down and subsequent volume change. Solid lines indicate direct relationships, whereas dashed lines are inferred or indirect associations. Solid boxes are hydrological parameters, whereas dashed are processes or actions xi

12 1. Introduction 1.1. Context Wetlands represent nearly 3% of the worlds land surface (Tarnocai, 1998) and are estimated to contribute approximately 12% of annual global methane emissions (Hansen et al., 1989), and hence are a significant component of the global carbon cycle (Moore et al., 1998). Over 14% of Canada s land area is wetland, representing 25% of the world s wetlands (Tarnocai, 1998). It is well established that hydrology is one of the most important overall controls on the carbon budget of wetlands (Moore et al., 1998). By their nature, the hydrology of wetlands (which are predominantly found in northern latitudes (Roulet et al., 1992)) is sensitive to changes in climate because of the delicate balance between evaporation and precipitation (Clair, 1998). Many global circulation models predict that northern latitudes will be subject to the greatest changes in temperature and precipitation under climate change conditions (Mitchell, 1989). Therefore, an understanding of the hydrological changes that may occur under a warming scenario is paramount in being able to understand the role wetlands play in global biogeochemical cycling. Roulet et al. (1992) modeled the hydrological response of a 2 x CO 2 climate scenario (increase in temperature and precipitation of 3 C and 1 mm d -1, respectively (Mitchell, 1989)) and predicted a ~14 22 cm draw-down in the water table. This was then applied to a subarctic/northern boreal wetland system (Moore et al., 1990) to evaluate the role climate change may play in carbon gas (particularly methane) dynamics. Some studies 1

13 suggest that under warmer conditions wetlands will act as a source of CO 2 and a sink for CH 4 (Blodau and Moore, 2003; Waddington and Price, 2000); a reversal of their current role. However, Roulet et al. s (1992) approach was simplistic because modifications of the wetland structure, and the consequent hydrological response were not considered. Therefore, an evaluation of the nature and magnitude of hydraulic change is needed, as are the implications on the hydrological regime of a wetland system. In Canada, wetlands are defined as land that is saturated with water long enough to promote wetland or aquatic processes as indicated by poorly drained soils, hydrophytic vegetation and various kinds of biological activity which are adapted to a wet environment (National Wetlands Working Group, 1997). Wetlands are subdivided into two broad categories: organic and mineral wetlands (National Wetlands Working Group, 1997). Organic wetlands (commonly called peatlands) represent over 90% of the wetlands in Canada (Tarnocai, 1998). The focus of this study is a peatland system. Peatlands are subdivided into fens, bogs and some swamps (swamps will not be addressed in this thesis). Bogs are ombrogenous (generally only receiving water inputs through precipitation), whereas fens can receive some surface and subsurface water input (National Wetlands Working Group, 1997). Fens (which are the primary focus of this thesis) are further classified into 19 different types (National Wetlands Working Group, 1997); one of particular interest is the String Fen. String fens are a type of patterned ground which is a prominent landform in boreal and subarctic regions (Foster and King, 1984; Price and Maloney, 1994; Quinton and Roulet, 1998). Despite the prominence of patterned peatlands, relatively little is know about both the origin and hydrology of these systems, especially how they may respond to climate change. 2

14 The Kyoto protocol is an extension to the commitment of the United Nations Framework Convention on Climate Change (UNFCCC) and calls for a decrease in greenhouse gas emissions and an accounting of sources and sinks of carbon as a direct result of anthropogenic land use change (limited to afforestation, reforestation, and deforestation). Despite Kyoto s bias towards anthropogenic land use changes, Roulet (2000) states that if the objective of the UNFCCC is taken literally, then all sources and sinks, regardless of origin, should be accounted for. Because a substantial portion of the world s wetlands are located in Canada, (which are known sources and sinks of atmospheric carbon), understanding how these systems will respond in warming climates becomes increasingly important. By studying the affects of water table draw-down in fen peatlands, a better understanding of the hydrological changes that may occur under a warming scenario will be attained, which is paramount in being able to understand peatlands future role in global biogeochemical cycling and carbon storage Patterned Peatlands: Formation and Function Formation Peatlands represent a long-term sink for carbon the accumulated remains of incompletely decomposed plant materials in the wet anoxic environment (Clymo, 1983). In some circumstances peatlands evolve with a distinct pattern of alternating pools and ridges (Foster and King, 1984). The formation of patterned peatlands is complex. Foster et al. (1983) propose a hypothesis whereby drainage is impeded at the base of a gentle slope. This impediment could be from an ice push ridge that prevented flow into the lake, 3

15 accented by melt water from the deeper snow found on the lee side of trees (which grow on the ice push ridges) found along the shoreline, creating waterlogged conditions (Foster and King, 1984; Foster et al., 1983). In these wet areas peat forming vegetation, such as Magnocaricetum, may begin to colonize (Foster and Fritz, 1987). As the peat develops it helps create a more homogenous surface which favours sheet flow (opposed to channelled flow) (Foster et al., 1983). The newly formed peat tends to have a lower hydraulic conductivity (K) (Foster et al., 1983) which further helps to impede drainage. Hydraulic conductivity governs the rate at which a liquid (e.g., water) can flow through a porous medium (e.g., peat) for a given energy gradient (Freeze and Cherry, 1979). Water that collects in any topographically low surface areas creates small ponds. The vegetation (e.g., Carex exilis, Scirpus cespitosus, and Sphagnum) that borders these small ponds creates a hummock and hollow microtopography (Foster and King, 1984; Foster et al., 1983). The hollows, because they are inundated, have poor peat producing vegetation, thus the contrast in elevation between hummock (with greater peat accumulation) and hollow becomes larger and the pools deepen and become more defined. Once the pools are able to maintain significant standing water, decomposition of the pool bottom will begin, further deepening the pool bottom (Foster et al., 1983). During wet periods the depressions (or pools) may begin to coalesce laterally (along a contour), creating larger, but narrow pools perpendicular to the slope. Price and Maloney (1994) found that pools in a patterned fen were typically roughly oblong in shape, m long and 5-7 m wide, with ridges 0.1 to 0.4 m high. As a ridge continues to grow, it will impede drainage from upslope, allowing the process to repeat itself. 4

16 This process of turning land into wetland is called paludification. Foster and King (1984) and Foster and Fritz (1987) conducted similar experiments in Leech Lake Peatland, Labrador, Canada and in patterned fens in Dalarna, Sweden. In both Sweden and Labrador, the ages (confirmed with radio carbon dating) of the pools decreased with increasing elevation, supporting the theory that the initial pools were formed at the base of the slopes (Foster and Fritz, 1987; Foster and King, 1984; Foster et al., 1988) and that paludification is important. Price and Maloney (1994) also found that dominant flow paths ran perpendicular to the alignment of pool/ridge topography Hydrology Despite the attention that has been given to understanding the formation of patterned peatlands (Foster and Fritz, 1987; Foster and King, 1984; Foster et al., 1983; Foster et al., 1988) relatively few studies have tried to understand and quantify the hydrology of patterned peatlands (Price and Maloney, 1994; Quinton and Roulet, 1998). Fortunately, the hydrology of any wetland system can be described using the water balance concept, which is essentially an accounting system where water is the currency (Ingram, 1983). A typical water balance equation for a peatland is; P + SWI Et Q Qss = S + ξ, Equation 1.1 where P, is precipitation, SW I is surface water inflow, E t evapotranspiration, Q is the surface discharge, Q ss is the subsurface discharge, S is the change in storage and, ξ, is the residual term (Equation 1.1 modified from Price and Maloney, 1994). The dimension of the previous terms is length [L], usually expressed as a depth in mm. 5

17 Physical Properties and Structure of Peatlands The topographical features (hummocks/ridges and hollows) that result in patterned peatland formation are also the same features that control the current day hydrological interaction between and within pools, and the subsequent carbon cycling (Belyea and Clymo, 2001). The presence of alternating layers of variably degraded peat, and sequences of vegetation within the vertical profile of most peatlands (Siegel, 1983), is evidence that this process is part of the normal evolution of peat systems. Ingram (1978) identified two distinct layers within a peatland, called the acrotelm and the catotelm. The upper, acrotelm, is the variably saturated layer composed of living, dead and poorly decomposed mosses (Price et al., 2003) and its thickness is defined by the depth from surface to lowest water table position (Ingram, 1978), usually between 0 and 50 cm (Price et al., 2003). In the acrotelm the peat is generally of lower bulk density (mass of solids/total volume, typically <0.07 g cm -3 (Van Seters and Price, 2002)), higher porosity (volume of voids/total volume, typically >90% (Baird and Waldron, 2003)) and higher hydraulic conductivity (typically >10-3 cm s -1 (Rycroft et al., 1975)). Consequently, the volumetric moisture content (volume of water/total volume) is lower (however, it can be up to 95%). Volumetric moisture content rises quickly during precipitation events but drains relatively quickly too. Hydraulic conductivity in the acrotelm decreases with depth by more than 4 orders of magnitude over 50 cm (Hoag and Price, 1995). It is this difference in hydraulic conductivity that can regulate the amount of sub/surface and surface water outflow and thus infiltration of water into the catotelm (Rycroft et al., 1975). The lower, catotelm layer is saturated, has a higher bulk density (typically > 0.1 g cm -3 (Van Seters and Price, 2002)), lower porosity and lower hydraulic conductivity (typically <10-4 cm s -1 (Rycroft et al., 1975)) (Ingram, 1978). 6

18 Because of saturated conditions, the pore spaces are generally filled with water, therefore volumetric moisture content will equal porosity (with the exception of biogenic gas bubble formation (see Baird et al., 2004a; Beckwith and Baird, 2001; Kellner et al., 2005)). The hydraulic parameters noted above control the timing and magnitude of water fluxes and stores for given climate inputs Surface and Subsurface Flow in Patterned Peatlands Surface flow in patterned peatlands is strongly influenced by the antecedent conditions (Quinton and Roulet, 1998). Quinton and Roulet (1998) note two distinct hydrological phases: 1) when water supply exceeds the depression storage and 2) when seepage and evaporation exceed inputs and the pools become isolated. (This is similar to the lateral coalescence process mentioned earlier during larger pool development.) When the conditions are saturated (directly following spring melt) surface flow (SW I and Q) between pools occurs through the acrotelm across ridges and around ridge flanks. Price and Maloney (1994) found that the pool-ridge sequence of a patterned fen has a very large depression storage as the pools will be able to fill with water after precipitation events when there were dry antecedent conditions. Subsurface flow is often considered negligible for a few reasons. As noted previously, the slope in these systems is generally quite low; for instance, Quinton and Roulet (1998) found a slope of and Price and Maloney (1994) found a water table gradient of 0.006, this, combined with a lower hydraulic conductivity in the catotelm, will result in minimal subsurface flow (according to Darcy s Law). Price and Maloney (1994) found subsurface flow to be about 3% of daily total runoff. Another reason that Q ss is considered negligible 7

19 is that patterned peatlands can only form when drainage is impeded and water pools. If the deeper subsurface (catotelm) was of a high enough hydraulic conductivity to permit significant flow, patterned peatlands would not form under the paludification hypothesis discussed earlier. Not all water movement in peatlands is horizontal. In fact, the vertical processes in boreal peatlands can dominate the water balance during summer (Price, 1996). The location of the water table within the acrotelm has profound implications for both storage changes, as well as carbon cycling. Hoag and Price (1995) note that because of fluctuations in water table position, the acrotelm experiences much greater changes in storage than the catotelm, which experiences no water table variation, and thus experiences very little change in storage. Specific yield, S y, is the ratio of the volume of water yielded by gravity drainage to the volume of the block of soil (Price et al., 2003). Storage changes, S, are controlled by the magnitude of water table change, h, and the specific yield ( S = h*s y ) (Hoag and Price, 1995). However, the acrotelm tends to have a very high specific yield (typically increasing from 0.2 to 0.6 near the surface) because of the higher porosity of the acrotelm, whereas the catotelm has a lower specific yield (typically 0.2 to <0.06) (Price, 1996) and thus higher water retention capacity (Schlotzhauer and Price, 1999). This implies that in the acrotelm a large amount of water must be removed to lower the water table. Branfireun and Roulet (1998) found significant increases in water table position, yet minimal increases in discharge when there were dry antecedent conditions with a low water table, however, with wet antecedent conditions (and a high water table) discharge increased rapidly. As discussed earlier, the lower hydraulic conductivity of the catotelm prevents significant subsurface seepage. Also, during dry antecedent conditions the pools are isolated, which 8

20 means that the dominate water loss must be due to evapotranspiration (equation 1.1). Typically wet conditions are only found following significant precipitation events and during spring snow melt Evapotranspiration Evapotranspiration (E t ) combines evaporation (from open water pools) and transpiration (from vegetation lawns, ridges/hummocks). Thus, it is largely dependant on the proportion of open water bodies within the peatland and the vegetation type (after overhead climatological conditions are considered). Evapotranspiration is the dominant outflow component of the water balance (e, quation 1.1), outside of the snowmelt period (when SW O and Q can dominate). For instance, Price and Maloney (1994) found that, post snow melt, evapotranspiration accounted for 126 mm in a Labrador fen (when precipitation only totalled 120 mm) over a six week period. A common method of estimating evapotranspiration is the Priestley-Taylor combination model, which has been commonly adopted by wetland scientists for the estimation of evapotranspiration as it requires less intensive field instrumentation than other methods. The alpha, α, value (see Methods section 3.1) is a proportionality constant (parameter) that is the ratio of actual evapotranspiration (E a ) (determined empirically using lysimeters for in situ estimates) to equilibrium evapotranspiration (E eq ) (the amount of water that could be evaporated into an atmosphere with no vapour pressure deficit (VPD)). Determination of the alpha value requires independent estimation of E a for similar surfaces. This can be achieved with lysimeters. Lysimeters are soil cores placed in buckets which are then placed into the hole left from the soil core. Lysimeters actually yield discharge measurements, 9

21 which can then be used to estimate evapotranspiration (Kelemen and Ingram, 1999). Because VPD commonly occur in a peatland, the alpha value is often greater than 1.0. Price and Maloney (1994) found alpha values of 1.55 and 1.27 for a fen pool and ridge, respectively. In a similar location (north eastern Quebec/Labrador, Canada) Quinton and Roulet (1998) found comparable alpha values of 1.6 and 1.34 for pool and non-pool surfaces. Price (1997) found alpha to be 1.21 for a bog in south central Quebec, Canada. The use of lysimeters is inherently problematic, however, as matching the internal (inside the lysimeter bucket) and external (the soil surrounding the bucket) moisture condition is difficult (Kelemen and Ingram, 1999). Further, weighing errors can be large, and errors in estimating precipitation must be incorporated Intra-pool Hydrology Little research has been conducted that specifically looks at the local (pool ridge) scale hydrology. Price and Maloney (1994) found that there were evaporative differences between ridges and pools (0.5 mm d -1 ) and significant differences between fens and bogs, mainly as a result of depth to water table, as fen ridges tended to be lower and wetter than bog ridges. The ability of a system to sustain a water table that replicates the topographic profile is a function of the hydraulic conductivity for a given set of water inputs (Ingram, 1982). Microtopography (hummock/ridge and hollow) with sufficiently low K, can result in ground water ridges or mounds that can control the lateral flow direction (e.g., Price and Maloney, 1994). However, as noted previously (see Section and ) hydraulic conductivity is very important to the creation of patterned peatlands, and is in a symbiotic relationship with plant growth (as the plants grow in wetter, lower K areas, and subsequently decay, they are creating an increasing lower K). Kellner and Halldin (2002) 10

22 state that different moisture content dynamics in acrotelm peat, between ridges, hummocks and hollows, is dependant on the water retention properties of the peat in those topographic features. (While hummock and hollow topography is limited to bogs, lawns and mats in fens would subject to similar dependencies). Ridge species tend to be smaller and more densely packed, and subsequently are able to retain and transport water more effectively (in the acrotelm, as a result of a lower specific yield). Kellner and Halldin (2002) found that the thickness of the unsaturated zone (i.e., depth to water table) varied the most in the ridges. During drying periods, the response of groundwater levels in ridges and hollows was similar, however, the hollows responded greater to precipitation events (Kellner and Halldin, 2002). Despite this, water flow from hollow to hummock was small. While considerable research could be conducted examining the water balance of patterned peatlands and the subsequent hydrological processes, all could be significantly inaccurate unless the non-rigid nature of peat is considered Peat volume change Peat is not a rigid soil because of its high water content (Price and Schlotzhauer, 1999) and large compressibility, thus changes in water table position (seasonally or long term change) can alter the storativity of peat (Price, 2003; Price and Schlotzhauer, 1999; Schlotzhauer and Price, 1999). The seasonal effect has been termed mooratmung (German for mire breathing ), which describes the vertical movement of the peat surface (Ingram, 1983). Volume change in peat may occur by three processes related to a change in water table position: 1) compression, 2) shrinkage and 3) oxidation. Compression occurs as the weight of material overlaying a point in a peat matrix is transferred from the fluid to the soil 11

23 structure, which happens when the water pressure decreases (e.g., with a water table decline). When the water table is lowered, the peat structure becomes unable to support the overlying material and the pore structure collapses, resulting in compression of the peat matrix and a lowering of the surface. The force of the overlaying material (total stress, σ T ) is a product of the depth, h, and the total density, ρ T, of the material overlaying it, and the acceleration due to gravity, g; σ T = ρ T gh. Equation 1.2 Fluid pressure (or pore water pressure, ψ) provides a buoyant forces against the total stress. Thus effective stress, σ e, is the stress placed on the structure of the peat not borne by the fluid. Thus; σ e = σ T ψ. Equation 1.3 Changes in effective stress can help explain the amount of compression that occurs in peat (Price, 2003). Shrinkage occurs above the water table. Shrinkage is the contraction of the peat matrix resulting from the water tension within the soil pulling the peat together (Price and Schlotzhauer, 1999). Price and Schlotzhauer (1999) note that normal compression and shrinkage are at least partly reversible. Kennedy and Price (2004) found in cutover peat that shrinkage was nearly 60% of seasonal volume change, compression nearly 40% and the remainder was due to oxidation. Oxidation can lower the surface by breaking down (oxidizing) the peat soil in the (primarily) aerobic zone (the zone above the water table) reducing pore spaces. The reduction in pore space comes from the release of carbon (e.g., carbon dioxide gas, 12

24 methane gas, dissolved organic carbon runoff) and the remaining smaller particles becoming more tightly packed, which reduces the pore space and increases bulk density. Volume change due to oxidation is irreversible. Rates of oxidation in peatlands are not well understood, especially long term rates (Waddington and McNeill, 2002). Waddington and McNeill (2002), found that the long term and intermediate oxidation rates in a disturbed/cutover site were similar at 5.7 and 6.2 mm yr -1, respectively, and the contemporary rate was 4.8 mm yr -1. Waddington and McNeill (2002) conclude that hydrology (Price, 1997) and peat structure (Price, 2003; Price and Schlotzhauer, 1999; Schlotzhauer and Price, 1999) are the main controls on the long term oxidation rate (as they control the water table position) Effect of peat volume change on hydrological parameters Compression affects the main hydraulic parameters (see section 1.6) including: porosity, n, bulk density, ρ d, hydraulic conductivity, K, specific yield, S y, and volumetric moisture content, θ VMC. As the water table drops, compression causes the porosity to decrease as the larger pores collapse first (Chow et al., 1992). As the porosity decreases, bulk density must increase assuming the particle density ρ s, remains constant, porosity is given as; n ρ ρ d = 1. Equation 1.4 s The hydraulic conductivity also decreases with the collapsing of larger pores since the large pores conduct most of the flow (Chow et al., 1992). Specific yield will decrease as the more tightly packed particles retain a greater amount of capillary water in the smaller pore spaces (Price, 2003). The volume change processes in peat are a response to changes in the 13

25 water table position, caused by climate variability or other anthropogenic causes. Considerable hydrological research has been conducted in cut-over peatlands (e.g., Kennedy and Price, 2004; Kennedy and Price, 2005; LaRose et al., 1997; Price, 1996; Price, 1997; Price, 2003; Price et al., 2002; Price and Whitehead, 2001; Price and Whitehead, 2004; Schlotzhauer and Price, 1999; Van Seters and Price, 2002; Waddington and McNeill, 2002; Waddington et al., 2002) from which a great deal of insight can be gained regarding the nature and magnitude of drainage and peat volume change. Chow et al. (1992) found that porosity decreased by 7% (from 92 to 85%) when compressed, resulting in a increase in bulk density of nearly 100% (0.124 to g cm -3 ) (equation 4). Schlotzhauer and Price (1999) found that bulk density of cutover peat changed seasonally from 0.11 to 0.16 g cm -3 with changes in peat volume. With oxidation, Van Seters and Price (2002) found, over the longer term (~30 years), that bulk density increased 0.07 to 0.13 g cm -3 in a harvested site when compared to a nearby natural site. With compression, Price (2003) found decreases in hydraulic conductivity by up to 3 orders of magnitude with change in water table of ~40 cm, whereas Van Seters and Price (2002) found longer term changes of half an order of magnitude with oxidation. Van Seters and Price (2002) found specific yield declined by 50% as a result of volume change. Depending on how much compression has occurred, during periods of precipitation the peat can swell and experience an increase in volume, thus the changes to the hydrological parameters mentioned previously can reverse, although compression is not always fully reversible (Price, 2003). Kellner and Halldin (2002) found that 40% of storage changes in a Swedish bog could be explained by seasonal swelling and shrinking ( mooratmung ), while Price and 14

26 Schlotzhauer (1999) found it was about 70% in a partly restored cutover Quebec bog. Therefore, volume change directly affects the water flows and stores within the soil (Price, 2003), and hence geochemical exchanges from and within the peatland, as well as other biogeochemical processes (Strack et al., 2004). The understanding of peatlands response to various stressors is beginning to emerge, and models that integrate the complex array of processes (e.g., Kennedy and Price, 2005) can be used to provide better management planning for disturbed peatlands (Price et al., 2003), and to incorporate the important feedback mechanisms like those needed in global climate models (Letts et al., 2000). However, more field study is required to quantify the nature, direction and magnitude of peat soil hydraulic changes, particularly in response to water table lowering, and their implications on the hydrological regime. Consequently, the water table in a fen peatland near Quebec City was manipulated in 2002, and the hydrological response was measured. The objectives of this study are to determine how water table drawdown affects hydrological parameters and water exchanges in a patterned fen and bog. Specifically, the objectives are to determine 1) the effect of water table drawdown on the main hydrological parameters (n, ρ d, S y, θ VMC ); 2) specifically how water table drawdown affects hydraulic conductivity between mat, lawn and ridge topography; 3) how changes in these hydrological parameters affect water table position and variability in pool, mat, lawn and ridge topography; and 4) the implications for water flow and storage within and across pool systems. Finally, the implications of these changes will be considered from a climate change perspective. 15

27 2. Study Site 2.1. Overview The study area) is located 20 km east of Quebec City, near Saint-Charles-de-Bellechasse (46 75 N, W), Quebec, Canada (Figure 2.1). The site is a string fen (National Wetlands Working Group, 1997) remnant, surrounded by two actively vacuum harvested fields (north-east and south-east margins, an abandoned harvested field (north-west) and an access road (south-west margin). The remnant is approximately 120 by 220 m. Located within the fen are a series of pool systems which include the pool itself and the surrounding mat, lawn and ridge areas that were the focus of this study. Three pool systems were the focus of this study and are identified the Control, Experimental and Drained sites (Figure 2.2). The Control site water level was not manipulated, whereas the Experimental site was drained by approximately 20 cm on 11 June 2002 by a shallow hand-dug ditch connecting it to a pre-existing peripheral drainage canal. The Drained site was drained circa 1994 (approximately 8 years prior to the drainage of the Experimental site) by the landowners in preparation for peat harvesting (but subsequently was never harvested) (Strack et al., 2004). It was assumed that, pre-disturbances, all three pools were similar hydrologically. At the fen the dominant shrubs are Chamaedaphne calyculata, Kalmia angustifolia, Vaccinium angustifolia and Andromeda glaucophylla. The dominant sedges at the fen are Carex oligasperma, Carex limosa, Rhyncospora alba, Eriophorum virginicum and Scirpus subterminalis (in the pools). The mosses at the fen include: Hummocks: Sphagnum rubellum, S. papillosum, S. magellanicum; Lawns: S. magellanicum, S. fallax. Hollows: S. 16

28 majus, S. cuspidatum and also bare peat or a cover of liverworts (Cladopodiella fluitans and Gymnocolea inflata).the Drained site is similar but the hummocks have more bare peat and Polytrichum strictum moss. The general flow direction in the remnant is from Control Experimental Drained. The ridge in the Control site is ~25 cm higher than the ridge in the Experimental site, which is ~15 cm higher than the ridge of the Drained site. Within sites, the ridges were approximately 13, 28 and 20 cm higher than the lawns in the Control, Experimental and Drained sites, respectively. The Control site pond is the largest, followed by Drained and Experimental site ponds (~800, 200, and 100 m 2, respectively). All three sites are underlain by a clay layer, which is 80, 110, and 130 cm below the surface in the mat areas in the Drained, Experimental and Control sites, respectively. The climate for Quebec City (18 km north west of the site) is classified as a Moist Mid- Latitude Climate with Cold Winters (Koppen classification: Dfb). The average annual temperature for Quebec City is 4.0 C with average January and July temperatures of and 19.2 C, respectively (Environment Canada, 2005). Mean annual precipitation is 1230 mm with 26% falling as snow. 17

29 Figure 2.1 Ariel view of the Fen study site 18

30 Figure 2.2 Site photos taken June 8, 2004 of the Control (top), Experimental (middle) and Drained (bottom) sites. 19

31 2.2. Nomenclature The site was studied from May to September The pools in the Fen will be called Control, Experimental and Drained with the initials C, E, and D, respectively. This thesis was part of a PERG (Peatland Ecology Research Group) project involving McMaster University (primarily concerned with gas fluxes), Universite Laval (ecology and plant succession), Environment Canada (Dissolved Organic Carbon), and the University of Waterloo (hydrology). Thus, while some idiosyncrasies (which will become apparent) exist in naming conventions of instrumentation, the actual site names have been used in efforts to make this thesis usable by others involved with this PERG project. The author of this thesis was present for the 2003 and 2004 field seasons. 20

32 3. Methods Instruments marked with an asterisk (*) were connected to a Campbell Scientific Data Logger (either CR10, CR10x, 21x or 23x) and logged at 60 second intervals with outputs (e.g., average, total) every 20 minutes. The reader is directed to Figure Micro meteorological instrumentation A meteorological station was installed between the Control and Experimental pools. Instrumentation at this site was replicated from , except were noted. Air temperature was recorded by using a copper-constantan thermocouple* located approximately 100 cm from the ground surface in a Styrofoam cup covered in aluminium foil. A soil temperature profile also used copper-constantan thermocouples* located at 0, 2, 5, 10, 20, 30, 50, and 70 cm below the surface. A tipping bucket rain gauge* was used to automatically record precipitation events. A manual rain gauge was located beside the tipping bucket and used as a data check. A photosynthetically active radiation (PAR) sensor*, and a net radiometer* were installed. In 2004, a second net radiometer* was installed over the Control pool. Two ground flux heat plates* were installed in the 2004 season. Evapotranspiration estimates were made using the Priestley and Taylor (1972) combination formula. Where s E = α s q Lρ * ( Q Q ) 3 G *10, Equation 3.1 and where s is the slope of the saturation vapour pressure temperature curve (kpa o C -1 ); 21

33 4098* Es s =. Equation ( T ) Es is the saturation vapour pressure (kpa); 17.3 T Es =, Equation where L is the latent heat of vaporization (J kg -1 ); ( * T )* 1000 L =, Equation 3.4 q is the psychrometric constant (assumed to be kpa o C -1 ), T is temperature in o C, α, is the ratio of actual and equilibrium evapotranspiration, Q* and Q G are the net radiation and net ground heat flux (J day -1 ), respectively, and, ρ, is the density of water (assumed to be 1000 kg m -3 ). The α coefficient, which represents the slope of the actual versus equilibrium evaporation relationship, was estimated using plastic lysimeters (see next section). 22

34 Figure 3.1 Site map of instrumentation 23

35 3.2. Lysimeters Lysimeters were used to estimate evaporative losses on the basis of mass changes due to precipitation/evapotranspiration, to and from, the peat (Kelemen and Ingram, 1999; Van Seters and Price, 2002). The lysimeters were constructed from plastic containers, either circular 20 litre paint buckets or rectangular Rubbermaid buckets. A bucket was perforated at the bottom, and placed into another, non perforated bucket of identical dimensions. Peat monoliths were placed into the perforated container. The perforated bottom allowed water to drain through the sample so that water content characteristics could be manipulated to match the conditions outside the lysimeter, and drainage into the non-perforated bucket could be subsequently measured. Lysimeters were weighed twice a week in 2003, and approximately five times a week in At the time of each weighing a qualitative inspection was completed to assess the needs of water content manipulation (removing water if soil surrounding looked drier by emptying the non-perforated bucket, or adding water if surrounding soil appeared saturated). Approximately once a week in 2004 a Hydrosense was used, as a quantitative check, to compare volumetric water content (VWC) between the surrounding soil and within the container. In 2002 three lysimeters were installed in the lawn area of the Experimental pool. In 2003 two lysimeters were installed in the mat and three installed in the ridge of the Experimental pool. In 2004 three nests of three lysimeters were installed in the mat, lawn and ridge of the Experimental pool. Two lysimeters were installed in the mat of the Control Pool. All lysimeters remained in the ground throughout the winter and were re-used the subsequent season. 24

36 3.3. Drainage A small ditch was constructed in June 2002 to facilitate drainage of the Experimental pool. This ditch extended from the drainage network of the abandoned peat field (north margin) to 3 m from the northern tip of the Experimental pool. A 3 m long, 10 cm diameter PVC (polyvinylchloride) tube connected the final 3 m from the ditch to the Experimental pool. The middle of the tube was buried approximately 15 cm below the original peat surface (leaving the two ends exposed). The pipe allowed the Experimental pool site s water level to drain, and be maintained at ~20 cm below the antecedent level Cores In August 2002 a Wardenaar corer was used to extract 3 cores in the lawn area of each site (Control, Experimental, and Drained). The cores were cut into approximately 4 equal sections (of roughly 15 cm in length) with depths centered at 15, 30, 45 cm, and, where possible, 60 cm. Standard methods (e.g., Freeze and Cherry, 1979) were used to calculate bulk density (ρ d ) and specific yield (S y ) Soil Moisture Content In 2004 Campbell Scientific Water Content probes* (CS 615) were in the mat/lawn areas of Control, Experimental and Drained sites at four depths (10, 20, 30 and 40 cm, respectively). 25

37 3.6. Squishometers Lines of elevation sensor rods (Price, 2003), henceforth known as squishometers, were installed in 4 locations in the 2002 season. At each of the pools, a sight wire (to provide a stable point of reference) was strung between two rebar poles (approximately 3-5 m apart) driven into the clay substrate below. The rebar poles were surrounded by PVC tubing to reduce the binding of peat onto the rebar (which could influence peat movement). The sight wires ran parallel with the edge of the pool (to sample similar topography and surface type). Squishometers were installed adjacent to the wire at various depths using two different anchoring techniques. For shallower squishometers (installed at less than 25 cm), a drywall screw was affixed to the end of a wooden doweling (with a diameter of 0.48 cm) (Figure 3.2). For deeper squishometers, spring loaded toggle bolts were affixed to the end of the doweling using a shrink-wrap tube. The squishometers were installed so that the doweling that protruded from the ground was very close to the site wire. Attached to this section of doweling was a small length of measuring tape (c. 20 cm). To determine vertical movement of the peat, the markings on the measuring tape were read against the sight wire. It was thought, a priori, that the most significant peat volume change would occur in the lawn areas, thus this is where the squishometers were installed. Thus, while this represents only a small portion of the site (~5%) it is the most important (arguably) for ridge pool water exchanges. At the Control pool, squishometers were installed to depths of 130, 100, 70, 50, 30, 20, 10 and 5 cm. Two sight wire lines were installed in the Experimental pool (called Experimental North and South (see Figure 3.2)) with squishometers in the North installed at 135, 100, 70, 50, 30, 20, 10, and 5 cm and in the South at 115, 70, 50, 30, 20, 10, and 5 cm. Squishometers installed at the Drained site were at 85, 50, 30, 20, 10, and 5 26

38 cm. The squishometers were read approximately twice a week in 2002 and 2003, and approximately five times a week in Figure 3.2 Top: Shallower (5 cm) squishometer (upper) and deeper squishometer (lower) prior to installation; Bottom: Squishometers installed at the Experimental South site (note use of white measuring tape, opposed to yellow). 27

39 3.7. Piezometer Installations and Locations All piezometers were made of polyvinylchloride (PVC) pipes. The radii and length of slotted intakes for all the piezometers can be found in Appendix A. Water level measurements were taken manually one to two times a week in 2002 and 2004, and two to three times a week in Figure 3.1 shows the locations of all piezometer nests and wells at the Fen site. In addition to the three fen pools already mentioned, two further pools were also studied. Control 2 is located between Control and Experimental and the water table was not directly altered. Drained 2 is located south of the Drained pool, and was also not directly altered. The Drained to Drained 2 (DDZ) transect inset map shows individual piezometers, and because the piezometers within a nest are located relatively close together (10s of cm), it was not necessary to repeat this scale in the other inset maps. Nest topography, installation year and pipe depths are summarized in Table 3.1. In 2002 three, three nest transects were installed perpendicular to the pool edge in the Control, Experimental and Drained sites (depths of slotted intakes shown in Table 3.1) encompassing mat, lawn, and ridge topography. In addition to the nomenclature previously mentioned (see Nomenclature page 20) Z stands for pie Z ometer. Thus EZ1 25 would read Experimental Piezometer nest 1 depth 25 cm. In addition to every nest always containing a well (W = well), these transects also had a fourth well installed in a pool topographic location. 28

40 In 2003 a five-nest piezometer transect connecting Control 2 (another un-manipulated pool close by) to Experimental (CEZ) was installed. CEZ1 was located on the Control side, whereas CEZ5 was on the Experimental side. Another transect was installed between the Experimental and Drained pools. EDZ1 and EDZ3 (there was no EDZ2) were in one transect connecting the two pools, and EDZ4, EDZ6, and EDZ8 (there was no EDZ5 or EDZ7) were in another. Again, wells were installed within each transect, as well as between nests (EDW 2, EDW3a, EDW5, and EDW 7). EDW3a was installed after EDW3 and EDW4 was installed and labelled (hence the a). In 2004 a 6 nest, 3 piezometer transect was installed between the Control pool and Control 2. CCZ1 was located in the original Control pool. A similar transect was installed between the Drained pool and a neighbouring pool (one that contained more water and looked similar to the Experimental pool) called Drained 2. DDZ1 (Drained to Drained 2) was located within the original Drained pool. In 2002, in addition to the piezometers listed above, piezometers were installed parallel to the lines of squishometers (see section 3.6; not shown in Table 3.1; 4 piezometers can be seen in Figure 3.2 behind the squishometers). The piezometers were installed so that the slotted intakes were at similar depths to the squishometers: Control: 25, 40, 60, 85, and 100 cm. Experimental North (not shown in Figure 2) front row: 25, 40, 60, 85, and 100 cm, back row: 60, 85, 100, 25, and 40. Experimental South (not shown in figure 2): 25, 40, 60, and 85 cm. Drained (not shown): 60, 40, and 25 cm. 29

41 Topography 2004 K test schedule Year Depths (cm) EZ1 Mat , 50, 75, 100 EZ2 Lawn , 50, 75, 100, 125 EZ3 Ridge 1 DZ1 Mat , 50, 75 DZ2 Lawn 1 DZ3 Ridge 1 CZ1 Mat , 50, 75, 100, 125 CZ2 Lawn CZ3 Ridge 1 CEZ1 Mat , 75, 100 CEZ2 Lawn 1 CEZ3 Ridge 1 CEZ4 Lawn 1 CEZ5 Mat EDZ1 Lawn EDZ3 Ridge 1 EDZ4 Lawn EDZ6 Ridge 1 EDZ8 Lawn CCZ1 Mat CCZ2 Lawn 1 CCZ3 Ridge 1 CCZ4 Ridge 1 CCZ5 Lawn 1 CCZ6 Mat DDZ1 Mat 1 DDZ2 Lawn 1 DDZ3 Ridge 1 DDZ4 Ridge 1 DDZ5 Lawn 1 DDZ6 Mat Table 3.1 Piezometer nest details and 2004 K test schedule, where 1, 2 and 3 refer to weeks 1, 2 and 3, respectively 30

42 3.8. Hydraulic Conductivity (K) Bail tests (Hvorslev, 1951) were used to determine K for each piezometer. The values of K were caculated as outlined in Freeze and Cherry (1979) based on Hvorslev (1951): ( L / R) 2 r ln K = Equation 3.5 2LT 0 where, r, and, R, are the internal and external radii of the piezometer, L, is the length of the slotted intake, and T 0, is the basic lag time parameter, which is calculated from the head recovery curve of the bail or slug test. In 2002 and 2004 water was drawn out of the pipe using a flexible rubber tube. Changes in initial head ranged from 10 to 50 cm. In 2003 a pump was used to ensure a consistent change in head (10 cm). The rate of head recovery was then measured. In 2002 each squishometer piezometer (e.g., CZK, DZK, EZK) had weekly tests conducted on them. In 2003 each pool piezometer (excluding those piezometers next to the squishometers) had one K test conducted on it between June and August. In 2004, K tests were preformed on a three-week cycle which was repeated five times (to total 15 weeks), as indicated in Table 3.1. Week 1 all piezometers at the Fen were tested (excluding the squishometer piezometers). Week 2 most mat and lawn nests, as well as the squishometer piezometers were tested. Week 3 the mat and lawn piezometers, as well as the squishometers piezometers were tested. The mat and lawn nests were tested weekly to assess a K dependence on water table depth (e.g., to assess if a seasonal drying trend affects K). All other piezometers were tested at least 5 times. 31

43 3.9. Surface and Water Level Recorders Pool water levels were recorded with electrical potentiometer devices* attached to a floatpulley system anchored to the sediment. Ground surface elevation changes were recorded with a similar device but with counter-weights instead of floats. Manual measurements of the distance from the centre of the wheel to the water level or surface level, respectively, were taken weekly. Where non-pool water table locations are reported, a 10 cm diameter PVC well or metallic stove pipe was used (Figure 3.3). The locations of all water/surface level recorders is shown in Table 3.2. Figure 3.3 Float-pulley system measuring the Control lawn water table. 32

44 RDS (Remote Data System) wells were used in The RDS well was programmed to log every 20 minutes and were data downloaded weekly. 5 RDS (Remote Data Systems) wells were installed in the: ridge between Experimental and Drained near well EDZ3; lawn near DZ1; DDZ6 pool; ridge near CZ3; and CCZ6 pool; respectively. Category Short Name Year(s) used Surface Level Recorders C sur lawn all D sur lawn all E sur lawn all Water level recorders E water ridge all E water lawn 2002, 2003 C water lawn all D water ridge all C water ridge all DZ1 water lawn 2004 Pool water level recorders C pool all D pool all E pool all CCZ6 pool 2004 DDZ6 pool 2004 Table 3.2 Water and surface level recorder locations and years used. (sur = surface) 33

45 4. Results The results of this thesis are divided into three sections: 1) Meteorological, 2) Hydrological Parameters, and 3) Hydrology. The rationale is an attempt to keep the reader focused on a specific theme or set of processes. The Discussion section will integrate all of the Result sections Meteorological Meteorological conditions were monitored from JD 131, 129 and 129 to 301, 268 and 233 for 2002, 2003 and 2004, respectively. The meteorological conditions for the field seasons were different, as indicated by the monthly average, maximum and minimum temperature values (Table 4.1). The 2004 field season was the warmest season with only June below the 30 year mean. Both 2002 and 2003 saw below normal average temperatures for all months. With the exception of June, July and August 2002, all monthly averages were within ± 2 C of the 30 year mean. 34

46 Temperature Precipitation Average Max Min Total 2002 May June July August May June July August May June July August year* May June July August Table 4.1 Average, minimum and maximum temperature values and precipitation totals per month for 2002 to 2004 seasons as well as the 30 year running average (*Environment Canada, 2005 at Quebec Lesage International Airport. Located 20 km west of the study site) Total precipitation for the recorded study period between May and August for 2002, 2003, and 2004 were 334, 349, and 444 mm, respectively (Figure 4.1), which were all less than the 30 year mean of mm (Table 4.1). There were 10, 10, and 15 precipitation events greater than 10 mm that accounted for 66, 64 and 74% of the total rainfall for 2002, 2003 and 2004, respectively. The 2002 and 2003 field seasons were similar in that they both had a number of relatively long (> 1 week) dry periods, whereas 2004 saw very few, long dry periods (Figure 4.1). The frequency of dry periods was determined by finding the length of time (in hours) between rain events (ignoring precipitation events of less than 0.5 mm ) (Figure 4.2). Average daily evaporation loss (E a ) measured with the lysimeters were 2.8, 3.1 and 3.5 mm day -1 for 2002, 2003 and 2004 respectively (Table 4.2, Figure 4.1) at the Experimental site. Equilibrium evaporation (E eq ) was calculated with the Priestley and Taylor (1972) model 35

47 for identical periods (some lysimeter data were rejected when heavy rain flooded lysimeters), and the ratio was used to estimate daily evaporation (E a ) thus α = E a /E eq. The water deficit (P E) was calculated to be 10, 7 and 72 mm for 2002, 2003, and 2004, respectively. 36

48 2002 Flux (mm) Cumulative P - E Precipitation Data missing, see note Evaporation Flux (mm) Flux (mm) Julian Day Figure 4.1 Daily precipitation (positive black bars) and evaporation (negative black bars) for 2002 to 2004 with cumulative flux (P-E) (line). Evaporative data (meteorological station) was missing from JD in 2002 and thus the average daily loss was added to the cumulative line. Note different scale in negative direction for

49 Year E a (mm day -1 ) E eq (mm day -1 ) Total Rainfall (mm) Total Evaporation (mm) Alpha Water deficit (mm) Table 4.2 Actual (E a ) and Equilibrium (E q ) evaporation rates, and water deficit (P E) for 2002 to Number of consecutive 24 hour periods without precipitation Frequency (per season) Figure 4.2 Frequency of 24 hour precipitation free periods for 2002 to 2004 field seasons. (In efforts to conserve the scale of < 24 hours (i.e., 1 on the y axis) is omitted because there are many (> 200) short time periods between rain events.) 38

50 4.2. Hydrological Parameters Drainage Drainage of the Experimental pool by 20 cm lowered the mat, lawn and ridge surfaces 25.5, 13.3 and 6.8 cm, respectively, over the 2002 field season (Figure 4.3, Figure 4.4). This can be observed visually (Figure 4.3) as yellow electrical tape was affixed to the piezometer at the pre-drainage surface level. Water levels dropped ~20 cm in the mat and lawn immediately (~4 hours) following drainage, whereas the ridge water table took considerably longer to decrease (~14 days, Figure 4.4). Figure 4.3 Surface subsidence along the Experimental transect. Yellow tape indicates the predrainage surface level with respect to the side of the piezometer. 39